From 20:30 UTC to 21:30 UTC, 16% of the hypervisors of the Paris zone failed to resolve the monitoring service domain name.
Applications which had instances on these hypervisors have been redeployed automatically because the monitoring could not reach them (even though they were available).